The stroke disease is caused because of the cerebrovascular accident which does not allow the vessels to supply blood to the brain. Occurrence of stroke is due to the burst or blockage of the blood vessel. For efficient diagnosis of ischemic stroke, Computed Tomography (CT) images are used with life support devices. Segmentation is one of the methods through which ischemic stroke region get differentiated from healthy tissues in CT images. However, accurate segmentation of original CT images is not obtained in minimum response time and pattern recognition is not carried out. Our research aims to perform early detection using segmentation and extracts the trivial features through optimality for classifying the ischemic stroke CT images into stroke and non-stroke images. In addition, pattern recognition is carried out for differentiation between strokes and non-strokes model.